Sensitivity of wave predictions to growth and dissipation parameterizations

Changes in wave heights during the propagation toward the coastline are influenced by many environmental factors. Under the complex effect of environmental processes, wave heights change due to energy dissipation, and the waves break under certain conditions. Due to the importance of predicting the wave heights in coastal areas, improving the prediction performance of spectral wave models by incorporating better physical descriptions of wave growth and dissipation attracts the interest of researchers. In this study, we aimed to quantify the sensitivity of wave height predictions to changing growth and dissipation parameterizations in a spectral wave model SWAN Cycle III Version 41.31A. Spectral wave model SWAN is used to hindcast the significant wave heights measured between the dates October 17 and November 10, 2018, at Shkorpilovtsi in the western Black Sea. Model runs are executed on an unstructured mesh-based spectral model of the Black Sea. The model is forced by ERA5 wind fields provided by European Centre for Medium-Range Weather Forecast. ST6 physics package is considered in this study for sensitivity analysis. The performance improvement in the prediction of Black Sea wave heights provided by the default parameterization of ST6 physics in comparison to default model parameterizations of Komen, Janssen, Westhuysen is shown previously. Further investigation on the sensitivity of modeled wave heights to ST6 parametrization is conducted in this study. Chosen ST6 growth and decay parametrizations needs to be coupled with swell dissipation terms. SWAN model presents Ardhuin and Zieger swell dissipation models. Coupling with the swell models provided ten model runs in total which are to be compared with the in-situ measurements by using error metrics including correlation coefficient, root mean square error (rmse), and bias. Sensitivity analysis based on given model runs are performed to identify the contribution of each parameter to resultant wave heights. Sensitivity analysis indicated an 8-30% of underestimation in the highest waves for the study area.

 

Acknowledgments: This research is supported by The Scientific and Technological Research Council of Turkey (TÜBİTAK) under grant number 119N400 and RFBR 20-55-46005.

 Keywords: Spectral wave modeling, wave hindcasting, sensitivity analysis